Self‐organizing tree‐growing network for the classification of protein sequences

Abstract

The self-organizing tree algorithm (SOTA) was recently introduced to construct phylogenetic trees from biological sequences, based on the principles of Kohonen's self-organizing maps and on Fritzke's growing cell structures. SOTA is designed in such a way that the generation of new nodes can be stopped when the sequences assigned to a node are already above a certain similarity threshold. In this way a phylogenetic tree resolved at a high taxonomic level can be obtained. This capability is especially useful to classify sets of diversified sequences. SOTA was originally designed to analyze pre-aligned sequences. It is now adapted to be able to analyze patterns associated to the frequency of residues along a sequence, such as protein dipeptide composition and other n-gram compositions. In this work we show that the algorithm applied to these data is able to not only successfully construct phylogenetic trees of protein families, such as cytochrome c, triosephophate isomerase, and hemoglobin alpha chains, but also classify very diversified sequence data sets, such as a mixture of interleukins and their receptors.Keywords: amino acid sequences; classification; neural network; phylogenetic reconstruction; self-organizing maps Neural networks (NNs) have several unique features and advantages over conventional statistical methods: they incorporate both positive and negative information; they are able to detect secondand higher-order correlation in patterns and a preconceived model is not required. These features make them particularly suitable for molecular sequence analysis. Since NN methods were first introduced in the analysis of sequence data to distinguish ribosomal binding sites from nonbinding sites (Storm0 et al., 1982), these techniques have found their applications in various fields of sequence analysis, including DNA introdexon discrimination and gene identification, DNA and protein pattern analysis, protein secondary and tertiary structures prediction, protein family classification, and phylogenetic analysis (for a recent review, see Wu, 1997).Neural networks may be classified as supervised or unsupervised according to their learning algorithms. A supervised network is trained by a data set of predefined organization scheme (e.g., a database organized according to family relationships), and used to classify new sequences into the data set. An unsupervised network, on the other hand, defines its own organization scheme according

Transcriptome Analysis Suggests That Chromosome Introgression Fragments from Sea Island Cotton (Gossypium barbadense) Increase Fiber Strength in Upland Cotton (Gossypium hirsutum)

“…We calculated the Pearson correlation coefficient (PCC) for all genes. We used the “Self-Organizing Tree Algorithm” (SOTA; MeV 4.9.0; ( Wang et al 1998 ) to cluster the log2-transformed RPKM values of all genes at 15, 20, 25, and 28 DPA. We clustered CSSL genes in CSSLs at 15, 20, 25, and 28 DPA with Cluster3.0 ( http://bonsai.hgc.jp/∼mdehoon/software/cluster/software.htm ; ( de Hoon et al 2004 ).…”

A New Method for Species Identification via Protein-Coding and Non-Coding DNA Barcodes by Combining Machine Learning with Bioinformatic Methods

“…It has been applied successfully in many fields, including speech synthesis, handwriting recognition and medical diagnostics. In molecular genetics it has been applied to some aspects of DNA/RNA and protein sequence analysis [51] , [52] , such as protein and ribosomal RNA classification [53] – [55] and phylogenetic reconstruction [56] . Some machine learning techniques have also been proposed for the analysis of DNA sequences, including Classification and Regression Trees (CART) [57] , [58] , Random Forest (RF) [58] , [59] , and Support Vector Machines [60] .…”